Multisamlet

Multisamlet is a conceptual construct used to represent multiple samples within a single analytical object. It is designed to facilitate cross-sample comparisons, cooperative modeling, and the capture of both shared and sample-specific information across related data.

A multisamlet typically comprises: a set of samlets (the individual samples), a feature matrix describing measurements across features for each samlet, and a relationship map indicating sample groupings, hierarchies, or lineage relationships. The object may also store metadata, normalization parameters, and a shared latent space produced by joint modeling techniques.

Analytical workflows using multisamlets include joint normalization, differential analysis across samples, clustering that respects cross-sample relations, and joint dimensionality reduction. Some proposed implementations support modular operations so that a multisamlet can be expanded with additional samlets without restructuring existing data.

Applications span genomics, transcriptomics, proteomics, and other high-dimensional data domains requiring integrated analysis across multiple samples. They are also discussed in theoretical contexts as a way to formalize multi-sample data fusion and to enable scalable analyses.

Origin and status: The term multisamlet appears in limited theoretical literature and in prototype software; it is not a widely standardized term and its definitions vary by context. The concept emphasizes modularity, reusability, and explicit representation of sample relationships.

See also: data integration; multi-omics; sample relationship networks.